Medical image fusion using guided filter random walks and spatial frequency in framelet domain

Abstract

The emergence of multimodal medical imaging technology greatly increases the accuracy of clinical diagnosis and medical analysis. Nevertheless, each medical imaging modal unavoidably owns its inherent limitations, so the fusion of multimodal medical images becomes an effective solution. In this paper, a novel fusion method on the multimodal medical images exploiting guided filter random walks and spatial frequency is proposed. First, the images to be fused are decomposed into a series of sub-images via framelet transform (FT). Second, a novel model called guided filter random walks (GFRW) is presented, which combines the advantages of both guided filter and random walks. Third, GFRW and spatial frequency (SF) are used to produce the fused approximate and residual images at each scale, respectively. Finally, the final fused image is reconstructed by using inverse FT. Experimental results indicate that, the proposed fusion method performs well in terms of subjective visual observation as well as objective evaluation metrics. Moreover, compared with other representative methods, the superiorities of the proposed method are still obvious, which is helpful to provide reference for clinical diagnosis and treatment.